1 | /**************************************************************************** |
2 | ** |
3 | ** Copyright (C) 2016 The Qt Company Ltd. |
4 | ** Contact: https://www.qt.io/licensing/ |
5 | ** |
6 | ** This file is part of the QtGui module of the Qt Toolkit. |
7 | ** |
8 | ** $QT_BEGIN_LICENSE:LGPL$ |
9 | ** Commercial License Usage |
10 | ** Licensees holding valid commercial Qt licenses may use this file in |
11 | ** accordance with the commercial license agreement provided with the |
12 | ** Software or, alternatively, in accordance with the terms contained in |
13 | ** a written agreement between you and The Qt Company. For licensing terms |
14 | ** and conditions see https://www.qt.io/terms-conditions. For further |
15 | ** information use the contact form at https://www.qt.io/contact-us. |
16 | ** |
17 | ** GNU Lesser General Public License Usage |
18 | ** Alternatively, this file may be used under the terms of the GNU Lesser |
19 | ** General Public License version 3 as published by the Free Software |
20 | ** Foundation and appearing in the file LICENSE.LGPL3 included in the |
21 | ** packaging of this file. Please review the following information to |
22 | ** ensure the GNU Lesser General Public License version 3 requirements |
23 | ** will be met: https://www.gnu.org/licenses/lgpl-3.0.html. |
24 | ** |
25 | ** GNU General Public License Usage |
26 | ** Alternatively, this file may be used under the terms of the GNU |
27 | ** General Public License version 2.0 or (at your option) the GNU General |
28 | ** Public license version 3 or any later version approved by the KDE Free |
29 | ** Qt Foundation. The licenses are as published by the Free Software |
30 | ** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3 |
31 | ** included in the packaging of this file. Please review the following |
32 | ** information to ensure the GNU General Public License requirements will |
33 | ** be met: https://www.gnu.org/licenses/gpl-2.0.html and |
34 | ** https://www.gnu.org/licenses/gpl-3.0.html. |
35 | ** |
36 | ** $QT_END_LICENSE$ |
37 | ** |
38 | ****************************************************************************/ |
39 | |
40 | #include "qimagescale_p.h" |
41 | #include "qimage.h" |
42 | #include <private/qdrawhelper_x86_p.h> |
43 | #include <private/qsimd_p.h> |
44 | |
45 | #if QT_CONFIG(thread) && !defined(Q_OS_WASM) |
46 | #include "qsemaphore.h" |
47 | #include "qthreadpool.h" |
48 | #endif |
49 | |
50 | #if defined(QT_COMPILER_SUPPORTS_SSE4_1) |
51 | |
52 | QT_BEGIN_NAMESPACE |
53 | |
54 | using namespace QImageScale; |
55 | |
56 | template<typename T> |
57 | static inline void multithread_pixels_function(QImageScaleInfo *isi, int dh, const T &scaleSection) |
58 | { |
59 | #if QT_CONFIG(thread) && !defined(Q_OS_WASM) |
60 | int segments = (qsizetype(isi->sh) * isi->sw) / (1<<16); |
61 | segments = std::min(segments, dh); |
62 | QThreadPool *threadPool = QThreadPool::globalInstance(); |
63 | if (segments > 1 && threadPool && !threadPool->contains(QThread::currentThread())) { |
64 | QSemaphore semaphore; |
65 | int y = 0; |
66 | for (int i = 0; i < segments; ++i) { |
67 | int yn = (dh - y) / (segments - i); |
68 | threadPool->start([&, y, yn]() { |
69 | scaleSection(y, y + yn); |
70 | semaphore.release(1); |
71 | }); |
72 | y += yn; |
73 | } |
74 | semaphore.acquire(segments); |
75 | return; |
76 | } |
77 | #endif |
78 | scaleSection(0, dh); |
79 | } |
80 | |
81 | inline static __m128i Q_DECL_VECTORCALL |
82 | qt_qimageScaleAARGBA_helper(const unsigned int *pix, int xyap, int Cxy, int step, const __m128i vxyap, const __m128i vCxy) |
83 | { |
84 | __m128i vpix = _mm_cvtepu8_epi32(_mm_cvtsi32_si128(*pix)); |
85 | __m128i vx = _mm_mullo_epi32(vpix, vxyap); |
86 | int i; |
87 | for (i = (1 << 14) - xyap; i > Cxy; i -= Cxy) { |
88 | pix += step; |
89 | vpix = _mm_cvtepu8_epi32(_mm_cvtsi32_si128(*pix)); |
90 | vx = _mm_add_epi32(vx, _mm_mullo_epi32(vpix, vCxy)); |
91 | } |
92 | pix += step; |
93 | vpix = _mm_cvtepu8_epi32(_mm_cvtsi32_si128(*pix)); |
94 | vx = _mm_add_epi32(vx, _mm_mullo_epi32(vpix, _mm_set1_epi32(i))); |
95 | return vx; |
96 | } |
97 | |
98 | template<bool RGB> |
99 | void qt_qimageScaleAARGBA_up_x_down_y_sse4(QImageScaleInfo *isi, unsigned int *dest, |
100 | int dw, int dh, int dow, int sow) |
101 | { |
102 | const unsigned int **ypoints = isi->ypoints; |
103 | const int *xpoints = isi->xpoints; |
104 | const int *xapoints = isi->xapoints; |
105 | const int *yapoints = isi->yapoints; |
106 | |
107 | const __m128i v256 = _mm_set1_epi32(256); |
108 | |
109 | /* go through every scanline in the output buffer */ |
110 | auto scaleSection = [&] (int yStart, int yEnd) { |
111 | for (int y = yStart; y < yEnd; ++y) { |
112 | const int Cy = yapoints[y] >> 16; |
113 | const int yap = yapoints[y] & 0xffff; |
114 | const __m128i vCy = _mm_set1_epi32(Cy); |
115 | const __m128i vyap = _mm_set1_epi32(yap); |
116 | |
117 | unsigned int *dptr = dest + (y * dow); |
118 | for (int x = 0; x < dw; x++) { |
119 | const unsigned int *sptr = ypoints[y] + xpoints[x]; |
120 | __m128i vx = qt_qimageScaleAARGBA_helper(sptr, yap, Cy, sow, vyap, vCy); |
121 | |
122 | const int xap = xapoints[x]; |
123 | if (xap > 0) { |
124 | const __m128i vxap = _mm_set1_epi32(xap); |
125 | const __m128i vinvxap = _mm_sub_epi32(v256, vxap); |
126 | __m128i vr = qt_qimageScaleAARGBA_helper(sptr + 1, yap, Cy, sow, vyap, vCy); |
127 | |
128 | vx = _mm_mullo_epi32(vx, vinvxap); |
129 | vr = _mm_mullo_epi32(vr, vxap); |
130 | vx = _mm_add_epi32(vx, vr); |
131 | vx = _mm_srli_epi32(vx, 8); |
132 | } |
133 | vx = _mm_srli_epi32(vx, 14); |
134 | vx = _mm_packus_epi32(vx, vx); |
135 | vx = _mm_packus_epi16(vx, vx); |
136 | *dptr = _mm_cvtsi128_si32(vx); |
137 | if (RGB) |
138 | *dptr |= 0xff000000; |
139 | dptr++; |
140 | } |
141 | } |
142 | }; |
143 | multithread_pixels_function(isi, dh, scaleSection); |
144 | } |
145 | |
146 | template<bool RGB> |
147 | void qt_qimageScaleAARGBA_down_x_up_y_sse4(QImageScaleInfo *isi, unsigned int *dest, |
148 | int dw, int dh, int dow, int sow) |
149 | { |
150 | const unsigned int **ypoints = isi->ypoints; |
151 | int *xpoints = isi->xpoints; |
152 | int *xapoints = isi->xapoints; |
153 | int *yapoints = isi->yapoints; |
154 | |
155 | const __m128i v256 = _mm_set1_epi32(256); |
156 | |
157 | /* go through every scanline in the output buffer */ |
158 | auto scaleSection = [&] (int yStart, int yEnd) { |
159 | for (int y = yStart; y < yEnd; ++y) { |
160 | unsigned int *dptr = dest + (y * dow); |
161 | for (int x = 0; x < dw; x++) { |
162 | int Cx = xapoints[x] >> 16; |
163 | int xap = xapoints[x] & 0xffff; |
164 | const __m128i vCx = _mm_set1_epi32(Cx); |
165 | const __m128i vxap = _mm_set1_epi32(xap); |
166 | |
167 | const unsigned int *sptr = ypoints[y] + xpoints[x]; |
168 | __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx); |
169 | |
170 | int yap = yapoints[y]; |
171 | if (yap > 0) { |
172 | const __m128i vyap = _mm_set1_epi32(yap); |
173 | const __m128i vinvyap = _mm_sub_epi32(v256, vyap); |
174 | __m128i vr = qt_qimageScaleAARGBA_helper(sptr + sow, xap, Cx, 1, vxap, vCx); |
175 | |
176 | vx = _mm_mullo_epi32(vx, vinvyap); |
177 | vr = _mm_mullo_epi32(vr, vyap); |
178 | vx = _mm_add_epi32(vx, vr); |
179 | vx = _mm_srli_epi32(vx, 8); |
180 | } |
181 | vx = _mm_srli_epi32(vx, 14); |
182 | vx = _mm_packus_epi32(vx, vx); |
183 | vx = _mm_packus_epi16(vx, vx); |
184 | *dptr = _mm_cvtsi128_si32(vx); |
185 | if (RGB) |
186 | *dptr |= 0xff000000; |
187 | dptr++; |
188 | } |
189 | } |
190 | }; |
191 | multithread_pixels_function(isi, dh, scaleSection); |
192 | } |
193 | |
194 | template<bool RGB> |
195 | void qt_qimageScaleAARGBA_down_xy_sse4(QImageScaleInfo *isi, unsigned int *dest, |
196 | int dw, int dh, int dow, int sow) |
197 | { |
198 | const unsigned int **ypoints = isi->ypoints; |
199 | int *xpoints = isi->xpoints; |
200 | int *xapoints = isi->xapoints; |
201 | int *yapoints = isi->yapoints; |
202 | |
203 | auto scaleSection = [&] (int yStart, int yEnd) { |
204 | for (int y = yStart; y < yEnd; ++y) { |
205 | int Cy = yapoints[y] >> 16; |
206 | int yap = yapoints[y] & 0xffff; |
207 | const __m128i vCy = _mm_set1_epi32(Cy); |
208 | const __m128i vyap = _mm_set1_epi32(yap); |
209 | |
210 | unsigned int *dptr = dest + (y * dow); |
211 | for (int x = 0; x < dw; x++) { |
212 | const int Cx = xapoints[x] >> 16; |
213 | const int xap = xapoints[x] & 0xffff; |
214 | const __m128i vCx = _mm_set1_epi32(Cx); |
215 | const __m128i vxap = _mm_set1_epi32(xap); |
216 | |
217 | const unsigned int *sptr = ypoints[y] + xpoints[x]; |
218 | __m128i vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx); |
219 | __m128i vr = _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vyap); |
220 | |
221 | int j; |
222 | for (j = (1 << 14) - yap; j > Cy; j -= Cy) { |
223 | sptr += sow; |
224 | vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx); |
225 | vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), vCy)); |
226 | } |
227 | sptr += sow; |
228 | vx = qt_qimageScaleAARGBA_helper(sptr, xap, Cx, 1, vxap, vCx); |
229 | vr = _mm_add_epi32(vr, _mm_mullo_epi32(_mm_srli_epi32(vx, 4), _mm_set1_epi32(j))); |
230 | |
231 | vr = _mm_srli_epi32(vr, 24); |
232 | vr = _mm_packus_epi32(vr, _mm_setzero_si128()); |
233 | vr = _mm_packus_epi16(vr, _mm_setzero_si128()); |
234 | *dptr = _mm_cvtsi128_si32(vr); |
235 | if (RGB) |
236 | *dptr |= 0xff000000; |
237 | dptr++; |
238 | } |
239 | } |
240 | }; |
241 | multithread_pixels_function(isi, dh, scaleSection); |
242 | } |
243 | |
244 | template void qt_qimageScaleAARGBA_up_x_down_y_sse4<false>(QImageScaleInfo *isi, unsigned int *dest, |
245 | int dw, int dh, int dow, int sow); |
246 | |
247 | template void qt_qimageScaleAARGBA_up_x_down_y_sse4<true>(QImageScaleInfo *isi, unsigned int *dest, |
248 | int dw, int dh, int dow, int sow); |
249 | |
250 | template void qt_qimageScaleAARGBA_down_x_up_y_sse4<false>(QImageScaleInfo *isi, unsigned int *dest, |
251 | int dw, int dh, int dow, int sow); |
252 | |
253 | template void qt_qimageScaleAARGBA_down_x_up_y_sse4<true>(QImageScaleInfo *isi, unsigned int *dest, |
254 | int dw, int dh, int dow, int sow); |
255 | |
256 | template void qt_qimageScaleAARGBA_down_xy_sse4<false>(QImageScaleInfo *isi, unsigned int *dest, |
257 | int dw, int dh, int dow, int sow); |
258 | |
259 | template void qt_qimageScaleAARGBA_down_xy_sse4<true>(QImageScaleInfo *isi, unsigned int *dest, |
260 | int dw, int dh, int dow, int sow); |
261 | |
262 | QT_END_NAMESPACE |
263 | |
264 | #endif |
265 | |